Data and code from: Diverse relationships between amplitude and frequency in bird vocalizations
Data files
Apr 18, 2025 version files 34.79 GB
-
ampfreq_data.csv
2.93 MB
-
brms_outputs.R2.zip
34.77 GB
-
Prum_Jetz_trees_20230414.tre
23.30 MB
-
README.md
6.54 KB
Abstract
Animals use sounds to communicate in contexts that are crucial for survival and reproduction. One compelling feature of acoustic signals is that two main domains of variation — frequency and amplitude — can interact with each other for reasons both evolutionary and mechanistic, in ways that are framed by conflicting predictions. To explore this issue, we obtained amplitude-calibrated field recordings and assessed relationships between vocal frequency and amplitude in 53 species of birds. Our main findings are twofold. First, we identify considerable variation across species in their amplitude–frequency relationships, emerging as positive in 27 species, negative in 12 species, and not discernible in 14 species. This variation, as well as results from phylogenetic models, suggest that amplitude–frequency relationships in birds are not governed by any universal rule. Second, throughout our sample — and particularly in oscines (songbirds) — we find that ranges of frequency variation are, by and large, amplitude-dependent, with 28 of our 35 most intensively sampled bird species producing broad ranges of frequency at low amplitudes yet narrower ranges at high amplitudes. This finding is consistent with the idea that birds are constrained to produce high-amplitude songs only within the narrow ranges of frequencies that resonate most constructively within their vocal tracts — which, if upheld, could lead to intensity–complexity trade-offs at the song level.
https://doi.org/10.5061/dryad.dv41ns26r
2025-03-21
The files deposited here should allow anyone to rerun all analyses and figures in the article “Diverse relationships between amplitude and frequency in bird vocalizations”. All you have to do is download all files to your R working directory, unzip the zipped folders, and run code as you wish.
Contact: jcarnioteles@umass.edu
Contents
- Commented R codes (codes.R2.html): This HTML file contains a step-by-step guide for reproducing all analyses and figures in the article. Available on Zenodo.
- File that generates codes.html (codes.R2.Rmd): This R Markdown file is the raw version of the HTML file above. Available on Zenodo.
- Main dataset (ampfreq_data.csv): This CSV file contains the data that underlie our analyses and figures. It is a duplicate of the article’s electronic supplement file Data S1. See Supplementary Methods (SM) sections 1-2 for a detailed explanation on how we constructed this dataset. See also metadata below for column meanings.
- Phylogenetic trees (Prum_Jetz_trees_20230414.tre): This TRE file contains a sample of 50 complete, species-level bird phylogenetic trees. They were constructed by taking advantage of Jetz et al.’s (2012) species-level resolution and Prum et al.’s (2015) genomic backbone, using codes and approach by Cooney et al. (2017). See SM section 4.5 for more details.
- Model outputs (brms_outputs.R2.zip): This ZIP file contains 7,334 R objects with the output of Bayesian single- and multi-level models generated by the package brms in R. Our results are based on these outputs. See codes for how these files were generated.
- dB_peak_within_[species name].rds (91 files): outputs from models estimating the effect of continuous amplitude on peak frequency within each of 91 species
- dB_peak_within_prior.rds: output from prior predictive check for model above
- dB_peak_meta[1/2]_oscine_tree[tree number].rds (100 files): outputs from models estimating the meta-analytic mean effect of continuous amplitude across oscines using each of 50 trees and each of two sample cutoff criteria (1, restrictive; 2, permissive)
- dB_peak_meta_oscine_prior.rds: output from prior predictive check for model above
- dB_peak_meta[1/2]_suboscine_tree[tree number].rds (100 files): outputs from models estimating the meta-analytic mean effect of continuous amplitude across suboscines using each of 50 trees and each of two sample cutoff criteria (1, restrictive; 2, permissive)
- dB_peak_meta_suboscine_prior.rds: output from prior predictive check for model above
- binary_peak_within_[species name].rds (44 files): outputs from models estimating the effect of binarized amplitude on peak frequency variability within each of 44 species
- binary_peak_within_prior.rds: output from prior predictive check for model above
- binary_peak_meta[1/2]_oscine_tree[tree number].rds (100 files): outputs from models estimating the meta-analytic mean effect of binarized amplitude across oscines using each of 50 trees and each of two sample cutoff criteria (1, restrictive; 2, permissive)
- binary_peak_meta_oscine_prior.rds: output from prior predictive check for model above
- binary_peak_meta2_suboscine_tree[tree number].rds (50 files): outputs from models estimating the meta-analytic mean effect of binarized amplitude across suboscines using each of 50 trees and the permissive (2) sample cutoff criterion
- binary_peak_meta_suboscine_prior.rds: output from prior predictive check for model above
- dB_SA1within[species name].D[iteration number] (4,550 files): output from supplementary models estimating the effect of continuous amplitude while simulating random head orientations 50 times for each of 91 species
- binary_SA1within[species name].D[iteration number] (2,200 files): output from supplementary models estimating the effect of binarized amplitude while simulating random head orientations 50 times for each of 40 species
- dB_SA2within[species name] (6 files): output from supplementary models estimating the effect of continuous amplitude with individual ID as random factor for 6 species
- binary_SA2within[species name] (1 file): output from supplementary model estimating the effect of binarized amplitude with individual ID as random factor for 1 species
- dB_SA3within[species name] (57 files): output from supplementary models estimating the effect of continuous amplitude on fundamental frequency for 57 species
- binary_SA3within[species name] (29 files): output from supplementary models estimating the effect of binarized amplitude on fundamental frequency variability for 57 species
Metadata for the main dataset
| Column | Type | Description |
|---|---|---|
| songID | string | vocalization segment ID |
| pointID | string | data point ID |
| species | string | species name (Jetz et al. 2012) |
| individualID | string | bird individual ID; slashes denote uncertain identity |
| centered_LAeq | numeric | distance– and noise-corrected amplitude reading, in dB LAeq, centered around each species’ mean amplitude |
| peak_frequency | numeric | distance- and noise-corrected peak frequency, in kHz |
| peak_harmonic | integer | number of the harmonic that contains peak_frequency |
| fund_frequency | numeric | fundamental frequency, in kHz; equals peak_frequency divided by peak_harmonic |
| group | string | taxonomic group: Oscines, Suboscines, or Nonpasseriformes |
Missing data code: NA
Details on how this dataset was constructed can be found in Supplementary Methods sections 1–2 of the associated article.